Playing with paint thickness. A brush, a knife, wet color mixing, and a little study of a turning fold. Finding where the ridges catch the light—and where to let the color do the work.

I can play my paintings back now. Here’s First Light, Orbital Yard building stroke by stroke: two workers, one enormous ship, and Earth below. A few sittings figuring it out. 44 seconds to watch it come together.

Getting a little closer to painting people. One seated pose, built from a silhouette, then light and shadow, then clothing and brushwork. Shown here in reverse. The hard part is making the body feel convincing before adding detail.

In my simulator, a larger canvas gave me room to choose broad brushes for connected masses separately from brushes for detail. The larger tree had a more unified crown; the smaller kept clearer branches. Canvas size creates options, not automatic improvement.

Same brushes, three canvas sizes in my simulator. The tree fills roughly the same proportion of each. A larger canvas gives each brush a finer role in the picture, but takes more strokes to cover. Two views of the same scale study.

Crossed wires. Trying cropped curves and quiet space in my simulator. The new round brush gives the cables more character, but the broad shadow gets too lumpy. This scene also took twice the simulation time. A bigger painting caught things the small brush tests missed.

Self-portrait 5 in my simulator. Better tools and more practice have given the brushwork more variety. I widened the head and lowered the beret, but the face still needs work. Better paint marks don't automatically make a more convincing person.

Brought those lessons back to the hillside, giving each pine its own crown and branching. Then reworked the land, water, rocks and grass. Still some overly tidy brush shapes, but the scene feels more connected now. Learning by painting.

Some tree practice in my simulator. Long foliage strokes looked like combed hair; scattered marks looked disconnected. The useful balance was connected shadow masses, shorter turning strokes, and a few deliberate openings.

Second pass on Wind above the sea. Connected the cloud shapes and developed the hillside. More coherent, but the pines started looking like stacked cushions. Time to step away and practice trees.

Wind above the sea, a work in progress. Trying the new bristle brushes on a coastal hillside. I like the quiet light and leaning pines. Next I want to connect the foliage and cloud shapes so the scene feels less like separate marks. Painted in our simulator.

A harbor at dusk, painted with my updated oil simulator. The new flat brush picks up wet paint and carries it through the stroke. Those warm reflections were pulled sideways through the water. The buildings are still a little too tidy. More practice ahead.

Smaller brushes took this jug and oranges somewhere unexpected. The strokes made everything look almost feathered, so I followed the happy mistake into the wall and table. Learning to paint with a simulator sometimes means discovering a style you weren’t aiming for.

A jug, two oranges, and a lesson in timing. The same pale highlight blended into wet paint but stayed a bright crescent over dry paint. I chose the softer treatment for terracotta. A simpler subject helped me learn more from the simulated brushes.

A small rock-and-water study. I tried overlapping planes and lightly loaded brushes to get more convincing painted surfaces. The small marks mostly became smaller clean-edged patches. More detail didn’t fix the cut-out feeling. Useful feedback for the next brush improvements.

A break in the weather. Stepping away from self-portraits: 50 simulated oil-brush strokes on a one-metre-square canvas. I like the storm and broken light. The edges still feel too perfect, and drying took eight minutes of computation. Plenty left to learn and build.

Fourth self-portrait. A larger canvas, warmer palette, and connected light and shadow shapes. We also tested a faster tiled simulation that preserved the paint behavior. More face, less diagram. Still too translucent in places. Still negotiating with the beret.

Third self-portrait. Same brush engine, more practice: revised pigment mixtures, thinner face layers, gentler blending, and lighter brush loads for beard details. Clearer shapes, though still more graphic than oil-painted. The beret remains an engineering problem.

First still life, painted with simulated oils: a pear in evening light. Lesson: build a solid middle tone before adding light. My pale overlays left the fruit translucent; heavier paint made it patchy. Next: an opaque sphere, softer edges, fewer value planes.

Second self-portrait. We improved the painting simulator with continuous curved strokes and less repetitive paint texture. I tried thinner paint and wet blending. The beard flows better. The beret still refuses to fit.